[erlang-questions] Two beautiful programs - or web programming made easy

Joe Armstrong erlang@REDACTED
Tue Feb 15 08:45:14 CET 2011


Re security.

It seems to me you can either secure the pipe (ssh) or secure the things in
the
pipe (ie the content).

I'd prefer to secure the content since I can do this with my own code - rsa
in erlang
boils down to A^C mod N with Erlang and bignums this is trivial - I trust my
own code.

Now suppose I've gotten as far as decrypting and authenticating the
encrypted blog
of code, dare I *run* the code - this boils down to trust and is a
non-technical issue.

Do I trust the NSA not to have put a back-door in ssh? - no. Do I trust my
rsa code - yes.

People happly click on "install this code" "run this code" buttons because
they trust
the source.

If I still want to run the code anyway, despite the fact I mistrust the
source I can run it
on a sand boxed machine that is not connected to anything.

But again do I trust the USB memroy (or whatever) that I used to transfer
the program?

What If somebody installs a keyboard logger in the firmware of my keyboard?

It's all about trust

/Joe




2011/2/15 Frédéric Trottier-Hébert <fred.hebert@REDACTED>

>
>
> On 2011-02-14, at 18:38 PM, Ryan Zezeski wrote:
>
> >
> >
> > 2011/2/14 Frédéric Trottier-Hébert <fred.hebert@REDACTED>
> >
> > This has *nothing* to do with encryption, and everything to do with not
> understanding the potential threats of the web correctly. It is an
> application-level issue, much like XSS is. And it's pretty damn important.
> >
> >
> > Except when it isn't.
> >
> > You make a very impressive summarization of web security problems, but
> there are many apps where these problems go away because they simply, don't,
> matter.  For example, how about building a new interface to the appmon
> application using this with Raphael or Protovis?  Do I really need to worry
> about all this crap?  I'd argue, no.  If I'm using appmon it's behind a
> firewall and I trust the user to be competent.  Just like 'rm -f 'in unix.
>  Plus, who gives a sh*t (pardon my French) about a hijacked session when
> someone else just cracked your gateway because of default passwords.  My
> point being, there are so many doors, and this focuses on one.  Getting it
> right is very hard, and thus very costly, and that's why I think the first
> thing to ask is "Does this even need to be secure?"  Given enough time,
> anything can be had.
>
> I won't argue with this. You're right on this. Again I'll repeat myself by
> saying that as a general (and generic) pattern, the eval() in Joe's code
> worries me. Individual cases can be tested and proven safe on an individual
> basis without too much trouble.
> >
> > If nothing else Joe has stumbled upon a way to rapidly produce
> nice-looking and feature rich GUIs in Erlang, and yet it seems everyone has
> somehow missed this point and focused on security and support and whatnot.
>  I'm glad there are Joe's in the world, otherwise it would be a pretty
> boring place.
> >
> > Don't let perfect get in the way of good enough.  As Joe said, life is
> too [expletive deleted] short.
> >
> > -Ryan
>
> There is nothing wrong with what Joe is doing. There is nothing wrong with
> exploration. However, things already exist in Javascript and on the web in
> general and we shouldn't reinvent the wheel all the time because we might be
> coming from a different domain or we don't have the same experience as real
> web developers have.
>
>  You only have to look into what people are doing with node.js. Just look
> at this example:
> http://www.screentoaster.com/watch/stUE5XQEVMRFtXQ1xVXFlYVlZX/serrano_session_sharing
> These guys are doing session sharing over the browser (including mouse
> movement and everything).
> You can also look at the nodeknockout as a whole:
> http://nodeknockout.com/teams (some links are now dead). These include
> multiplayer browser games, chat systems, drawing applications, charts, etc.
> Dreamers and innovators are at work on the web already, they're just not
> necessarily doing it with Erlang.
>
> They are doing that kind of explorative nice-looking innovation we all find
> awesome (even if they likely don't do it 100% safe either).
>
> I'm sorry if I sounded patronising or just bossy around this discussion,
> it's just that there are fantastic ideas to be inspired from in what is done
> by these programmers all the time. They usually do it cleaner, safer and
> smarter. At least on the client side.
>
> If people can learn from Erlang when it comes to heavily concurrent and
> fault-tolerant applications, Erlang programmers can certainly learn from
> other domains when the developers working there are more specialised at what
> they do than we could be as a community in here.
>
>
> --
> Fred Hébert
> http://www.erlang-solutions.com
>
>


More information about the erlang-questions mailing list